Skip to main content

verify_bitand_reduction

Function verify_bitand_reduction 

Source
pub fn verify_bitand_reduction<F, C>(
    log_constraint_count: usize,
    eval_domain: &BinarySubspace<F>,
    operands: &[OperandClaims<'_, C::Elem>],
    channel: &mut C,
) -> Result<AndCheckOutput<C::Elem>, Error>
Expand description

Verifies the batched BitAnd check: A & B == C on every row.

This is the univariate-skip zerocheck of A(Z, X) * B(Z, X) - C(Z, X) == 0 for all rows (Z, X), where Z is the bit index within a 64-bit word and X is the row index. Its multilinear rounds are one sumcheck with the operand-column MLE-checks of operands; see rerand.

§Arguments

  • log_constraint_count: base-2 logarithm of the row count — the operand column length, which the prover zero-pads up to a power of two. The single-instance verifier passes ceil(log2(n)) for n AND constraints; the batched M4 verifier adds its log instance count, since one row there is an (instance, constraint) pair.
  • eval_domain: the univariate-skip domain, one dimension above the 64-bit word, already lifted to F. The caller passes it so it matches the shift reduction’s domain by construction.
  • operands: the multiplication reductions’ per-bit operand claims. They must be in the transcript before this runs, since it draws the challenge that folds them.
  • channel: the verifier channel that reads messages and redraws Fiat-Shamir challenges.

§Errors

Returns an error if any sumcheck round message or the final consistency check fails.